MULTIPLE FUNCTION MAGNET SYSTEMS FOR MAX IV
F. Bødker, C.E. Hansen, N. Hauge, E. Krauthammer, D. Kristoffersen, G. Nielsen, C.W. Ostenfeld,
C.G.T. Pedersen, Danfysik A/S, Taastrup, Denmark
Abstract Danfysik is currently producing 60 magnet systems
with discrete multipole functions integrated into yokes for
the bending achromats of the MAX IV 3 GeV storage
ring. The integration of up to 12 multipole magnets into
individual yoke structures enables a compact, low
emittance storage ring design where the elements are
aligned with high precision inside the magnet girders.
Testing of the compact magnet structures with yoke
lengths up to 3.3 m has been a challenge which required
development of new test equipment dedicated to this task.
INTRODUCTION
Danfysik is in the process of producing 20 of each of
the MAX-lab magnet girder types M1, M2 and U3. Fig. 1
shows a mechanical layout of one of the magnet girders,
as designed by MAX-lab [1]. The top and bottom magnet
yokes, including the poles for a combined function
bending magnet, are machined out of one single iron
block. Separate, higher order multipole magnets are
mounted into the yokes. The machining of the up to 3.3 m
long yokes out of solid iron blocks to tight tolerances
requires special attention and advanced machining
procedures. This process is now fully implemented and
the required tolerances have been achieved.
The dipole and quadrupole elements are magnetically
field mapped on a precision Hall probe measuring bench
with special attention to achieve the best possible position
accuracy. All multipoles are measured on a slow rotating
coil system developed for that purpose. Much effort has
been put into automation in order to increase the
reliability and reduce the measuring time. Testing of the
first M1, M2 and U3 magnet girders is now completed.
Figure 1: Mechanical layout for the U3 magnet girder.
PRODUCTION
Complete magnet girders of each of the three types
have been produced together with iron yokes for the first
10 M1 girders. A picture of a finished M1 magnet girder
is shown in Fig. 2. The main requirement on the yoke
machining is a ±0.02 mm tolerance limit which is also the
tolerance needed for the separately machined multipole
pieces. In order to meet this tolerance and in addition to
obtain magnets with uniform magnetic properties, the
yoke parts are all made out of one batch of low carbon
Armco steel and heat treated as part of the production.
Machining has been outsourced and is performed on a
large scale CNC mill especially adjusted and calibrated
for this task. To ensure the best possible precision the
CNC machine is dedicated to machining of the MAX-lab
girders during the whole production period. The high
precision and small tolerances are achieved by using an
iterative machining refinement process of rough
machining and heat treatment followed by fine machining
adjusted to 3D measurements.
After machining, 3D measurement campaigns,
consisting of approximately 1300 point measurements,
are performed for each top and bottom yoke. Results for
the produced parts have been evaluated in cooperation
with MAX-lab and the results were within specifications.
Special attention was paid to the pole face geometry of
the dipole integrated in each yoke. The mechanical
tolerances of ±0.02 mm on e.g. the pole face geometry are
referenced to the very large mating surfaces of the yoke
halves and reference surfaces at each yoke end.
Figure 2: Top and bottom parts of an M1 magnet girder.
Meeting the tolerances on the sextupole and octupole
magnets, which are kinematically mounted into precision
machined slots of the yokes, required a special functional
machining to avoid the build-up of the tolerances on
individual iron pieces.
A special coil winding concept was developed in order
to minimize the risk of internal water leaks, in particular
in the well hidden coils of the multipole magnets. This
concept omits the need for internal joints inside the girder
for the sextupole and octupole coils and it further
facilitates simplified and faster coil production including
large scale impregnation of multiple coils.
MOODB102 Proceedings of IPAC2013, Shanghai, China
ISBN 978-3-95450-122-9
34Co
py
rig
ht
c ©2
01
3b
yJA
Co
W—
ccC
rea
tive
Co
mm
on
sA
ttri
bu
tio
n3
.0(C
C-B
Y-3
.0)
07 Accelerator Technology and Main Systems
T09 Room-temperature Magnets
FIELD MAPPING WITH HALL PROBE
Hall Measurement Setup
Field mapping of the combined function dipoles is
performed through dedicated slots in the side yoke, as
seen in Fig. 3. For quadrupole magnets the field gradients
are measured at three longitudinal positions through small
holes in the side of the yoke. Our Hall probe is mounted
on a sled which rests on a long granite beam with an
overall flatness of 37 µm. The sled is moved along the z
direction by a servo motor with feedback from a laser, and
the probe is moved by stepper motors using the feedback
from linear encoders with an accuracy of 3µm.
Figure 3: Field mapping with a Hall probe mapper on a
granite bench with longitudinal axis along the beam axis.
The relations between bench and girder coordinate
systems were found by scanning the probe over magnetic
pins at mechanically known girder positions. Repeated
scans over the same pin found the same pin position
within 2.2 µm. We have seen position drift of less than 10
µm during a period of several days.
Accurate on-the-fly field measurements were possible
using a Hall probe measuring system made at Danfysik
based on a concept adapted from GSI [2]. The Hall
element is placed in a non-metallic keeper including a
temperature sensor and calibrated against field and
temperature to field accuracy better than 1 G. The probe
is supplied by a current supply with 30 ppm long term
stability.
Software routines for fully automated control and
logging of environmental parameters minimize user
interaction and hence reduce the risk of errors in the
relative complex field mapping. In addition, they ensure
efficient, failure safe and reproducible measurement
conditions for the entire production series.
Hall Measurements
The bending dipoles are to be field mapped along lines
transverse to the nominal orbit with 5mm spacing and
along the curved orbit with a transverse spacing of 1 mm,
resulting in a total of up to 7600 points. If this was
realized by step-and-go measurement, it would take more
than 10 hours. Instead the dipoles are mapped on-the-fly
on a rectangular grid, and the required data obtained by
interpolation. The resulting measurement time was only
1.5 h, minimizing errors due to thermal drift.
The raw field values of the measured grid are shown in
Fig. 4, with the soft dipole end [1] to the right. The map is
made by on-the-fly measurements along straight lines in
the z-direction with 0.25 mm between samples and 1 mm
line spacing. This grid forms the basis for interpolation on
transverse lines and along the orbit, as shown in Fig. 5.
Figure 4: Magnetic field map of an M2 dipole. Darker
color indicates higher field level.
Figure 5: Magnetic field on the main orbit trajectory
obtained by interpolation from the measured grid data.
Repeated interpolated field maps show that the results
are very stable with relative standard deviations of 7·10-5
on the field integral levels and deviations of only 2·10-5
on
obtained effective magnetic dipole lengths.
Step-and-go field measurements were made on a
transverse line in the center region of an M2 dipole with a
magnetic field gradient of 8.7 T/m, making the
measurement strongly position sensitive. In Fig. 6 these
results are compared with interpolated results obtained
from three repeated full field maps of the dipole. The
relative deviation is less than 1.6·10-4
, validating the
interpolation scheme using on-the-fly measurements.
Figure 6: Relative deviation between discrete step-and-go
measurements and interpolated field values on a
transverse line of a combined function bending magnet.
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
0100200300400500600700
B [
T]
S [mm]
Proceedings of IPAC2013, Shanghai, China MOODB102
07 Accelerator Technology and Main Systems
T09 Room-temperature Magnets
ISBN 978-3-95450-122-9
35 Co
py
rig
ht
c ©2
01
3b
yJA
Co
W—
ccC
rea
tive
Co
mm
on
sA
ttri
bu
tio
n3
.0(C
C-B
Y-3
.0)
HARMONIC COIL MEASUREMENTS
The challenges of harmonic coil measurements of small
aperture multipole magnets are discussed by Buzio in [3].
Several measurement techniques were considered,
including multilayer PCB coil measurements. Inspired by
work on Linac4 [4], we opted for a non-compensated
tangential coil design using a 40 turn Litz-wire at an
opening angle of 15° in order to obtain adequate
sensitivity for the higher harmonics. The Litz-wire is
placed in precision grooves machined into a support rod
with at a measuring radius of 10.7 mm, which is
optimized for the pole radius of 12.5 mm. A total of 13
harmonic coil segments are made, one for each individual
magnet, with 3 to 5 individual coils on each support rod.
The lack of compensating windings meant that accurate
mechanical rotation of the coil was required, which was
achieved by using high precision ball bearings. The
harmonic coil system is inserted from a girder end (see
Fig. 7) relying on mechanical positioning. By calibrating
each coil segment separately, we could determine the
individual phase differences between the coils, as well as
the integrated strength transfer function. The coil signal is
recorded via a Metrolab integrator using a LabVIEW
program adapted from GSI in Darmstadt, Germany.
The measurement program has been modified to contain
all relevant calibration and test values and to fully
automate the measurement series, including current pre-
cycling and measurements at a range of current values.
This is particularly important with the need for
continuous switching between many multipole magnets of
different types using different coil segments.
Figure 7: Rotating harmonic coil measuring system
mounted on an end section with inserted pick-up coil.
The stability of the harmonic coil system was evaluated
by making repeated measurements on two quadrupoles
and two octupoles at their nominal field strength. These
short term repeatability tests gave on average a standard
deviation of 0.4 units on the main field gradient strength
(unit=10-4
), a negligible change on the determined magnet
center position and 0.01 mrad variation on the magnet
rotation value. The variations on higher harmonic terms
between repeated measurements were in all cases below
0.1 units. A simple thermal test where the harmonic coil
was locally heated indicated quite modest thermal drifts
of -0.03 mrad/°C for the magnet rotation angle and 0.2
unit/°C on the main harmonic gradient strength.
Extended harmonic coil measurement campaigns have
been performed for the first M1, M2 and U3 magnet
girders at a range of current settings for each of the 9 to
13 built-in magnets spanning from corrector, quadrupole,
sextupole to octupole magnets. Measurements were
repeated after repositioning of the test jig and again after
disassembly of the top and bottom yokes. These repeated
test jig mountings were a challenge for the measurement
reproducibility. Variation on the magnetic center position
with disassembly of the test jig and magnet is typically
below 0.01 mm. The magnetic center was found to be
within 0.1 mm of the harmonic coil rotation center, which
is the combined limit on coil and magnet misalignments.
The obtained variation of measured main harmonic
gradient is for quadrupoles on average within ±2 units.
For higher order terms it decreases from 0.4 to 0.05 units
with increasing order of the harmonics. Generally, higher
harmonic repeatability is well within 1 unit.
Higher harmonics of the other measured multipole
magnets show similar trends. Besides the allowed terms,
they are dominated by the first two higher harmonics. The
remaining higher harmonics are typically below 1unit
which shows that the pole profile is well performing. Fig.
8 shows sextupole and octupole harmonics as a function
of excitation current for an M2 quadrupole. The octupole
terms are nearly constant while the sextupole terms show
some variation. It seems like the variation is a result of
different yoke flux return paths in combination with
excitation dependent permeabilities. Similar trends are
observed for the sextupole and octupole magnets.
Figure 8: Measured normal and skew sextupoles (B3, A3)
and octupoles (B4, A4) harmonics as a function of current
for a focusing quadrupole on the M2 girder.
CONCLUSION
The first M1, M2 and U3 magnet girders have been
produced and tested at Danfysik and found to fulfill both
magnetic and mechanical specifications. The measuring
systems perform very well with high stability and
accuracy and our results are in general in good agreement
with model calculations by MAX-lab.
REFERENCES
[1] M. Johansson et al., IPAC 2011, p. 2427.
[2] F. Klos, GSI Darmstadt, Germany, personal communica-
tion.
[3] M. Buzio, “Magnetic measurement of small-aperture
quadrupoles at CERN”, PSI Villigen, 15 April, 2010.
[4] M. Buzio, personal communication.
MOODB102 Proceedings of IPAC2013, Shanghai, China
ISBN 978-3-95450-122-9
36Co
py
rig
ht
c ©2
01
3b
yJA
Co
W—
ccC
rea
tive
Co
mm
on
sA
ttri
bu
tio
n3
.0(C
C-B
Y-3
.0)
07 Accelerator Technology and Main Systems
T09 Room-temperature Magnets
Top Related